Unreal Engine 5 = Game Changer

let's be honest here, for me i'm a beginner and i have a mobile game idea working on it.
most of both engine features i won't use it , and most indie people won't.

but the biggest announce for me is unreal license is free to use and incurs 5% royalties when you monetize gross revenues from that product exceed $1,000,000 USD. that the game change ...
you have full engine features + source code and its all free while in unity i have to pay 40$ per month for plus version to use Splash screen customization.
unity should change that most of us Indie developer try do our game in free time and most of us don't have budget and we like to use good stuff . and i know my game probably won't reach $1,000,000 USD revenues that's mean i won't pay anything for unreal which sound perfect ....

unity respond should be in there engine plan

 

IMHO RTX can accelerate a generic solution without having to actually raytrace everything like we try now, apparently we can program a new primitive for accelerating cube instead of triangle for example (I have seen an example with sphere), given that GPU don't like non coherent access but RTX hardware don't have a problem with it, my first guess is that it would specialize into a first pass to get "in order" "out of order" data. Visibility being such problem, just having a wide pass that remove 95% of the data would be great, ie a top level accelerating structure. It's probably good as a screen level acceleration structure, leaving medium size to be done by other technique.

For those interested in the technical aspect of that GI, here is the fortnite medium level AO structure tracing they have. MY guess is that they use that with a large VXGI and SSGI into a unified version powered by their compute rasterization.

Obviously the one big unknown is that compute rasterization.

 

What marketing gimmick ? Having a realtime GI solution (even if lightmass already was top tier) ? Having a renderer that you can just throw polys at it and handle them automatically ? (even if they already had HLOD and impostors) ?

GIving away millions of dollars worth of assets, games and free online services might be a gimmick, but I totally welcome it.

There was some weird Sony product placement through the whole reveal though.

 

Which is why there's 500 instances of the same statue there rather than 100 instances of 5 different statues.

And I guess why the girl had to go pushing through that crevice, so that the outside world assets could be loaded in time

 

That's the biggest game changer IMO, and the most difficult nut to crack for Unity, because it directly hits in the business model, which was literally Epic's master plan, I mean they have Fortnite billions and Chinese investors, it's not about some shiny tech that can be copied and will be outdated in 3 years anyway. I'll be watching with a bowl of popcorn how this will play out.

For me subscription is worth it, I mean it's actually cheap compared to what it was back in the day when first you had to pay for every platform (one time fee but total was like $4500) then it was like $245 A MONTH (yes) and people complain about $40/month with full features, with a revenue limit that's basically unreachable for most indies anyway. And if you have a business based on Unity, it's not like you can switch on a whim, when version changes are a pain, let alone abandoning all the investments and tools you made, things you learned. So yeah, for existing users staying with Unity is not really a choice regardless of anything, but for new ones, yeah...UE sounds like a no brainer currently to be honest, as much as I love Unity.

 

On the surface it definitely looks like a game changer but what people from the outside may not be aware of is that you can negotiate a custom license with Epic Games. Once you're at the point where you're making a million dollars or more you'll have most likely started the negotiating process with them meaning this only truly helps small indie developers.

 

I'd like to see unity make an actual game
imagine what they could accomplish with intimate knowledge of dots and srp

 

16 years after epic delivered visual scripting, unity buys an asset store solution and still charges for it.
I know I write inflammatory prose but is it a lie?

 

Wow, a year ago Hippocoder would have censored a comment like this.

 

C# is better than visual scripting
does anyone here seriously need visual scripting?

 

Yeah that's the problem too, they're not eating their own dog food, if you make a hit game it can be a huge revenue stream, of course it's not easy by any means, but if you have your own popular engine and you know it inside out, have contacts in the industry...definitely easier than for most.

And the demos... the only demo that was memorable for me was Adam, and clearly not just for me since it has like 31M views, and the cinematic demos felt fresh at that time. Of course that was before UE4 got into cinematic stuff with Megascans and raytracing, now no one cares about Adam and Heretic was forgotten in 5 minutes, see how quickly things move in this industry? Also notice how Epic has made the demo somewhat like a game, kind of obvious since it was made for a game console, but they still could've just made a boring cinematic.

Unity is still selling shovels for a gold rush that ended several years ago. Now there's huge competition even among engines, just look at how much engine stuff is widely available for free, even for indies and for C# coders there's Godot, Xenko, and new ones keep popping up. Obviously Unity has more support, Asset Store and many other things, but still, it's not early 2010s anymore.

 

Plenty of people want and need visual scripting. Despite you not wanting it, it's a very powerful tool to give to the art team rather than trying to teach them HPC# code and also very useful if you just want to hammer out really simple operations.

 

obviously unity is not perfect, and so is no engine out there. but i really really like the direction is heading. honestly i think they should put pixel perfect graphics and visual scripting on the back burner and really polish dots and ui elemets. then you can safely say unity would be a the best engine imo

 

remember unity online stuff?

well the epic one is free for all engine

AND it's battle tested with fortnite

And you automatically benefit from the existing user base of epic + fortnite

 

Visual scripting is a core part of DOTS going forward and has always been stated as such.

 

Traditional scripting languages and visual scripting languages target completely different people. If C# works for you that's awesome but it likewise means that you were never the target audience for visual scripting.

 

I'm not a fan of visual scripting either, but it definitely has a place and a purpose. I'd much rather that my level designers use that over having to write C# scripts, even if it's just for separation of concerns.

- - -

Personally, I like the lighting thing, but there's still visible latency, and by the time this is in our hands I'm kind of hoping that other similar solutions are also around to choose from. I've read that other games have already implemented this kind of thing for themselves, so seeing generally available implementations in engines can't be that far off anyway, right? That's not to devalue what's on show here, I'd love it in my game and I'm looking forward to getting my hands on it, whomever it comes from.

The model thing, though... that is super cool. I don't care for anyone's product loyalty, that is a potentially amazing tool. Not so much because I can put "billions" of polys in my scene, but because I think that could probably be built upon to get both great results and super smooth workflows. Eg: can this be used to just automate the LOD process to the point where it becomes both developer and pixel transparent? Because if someone gives me a 30 million poly model to put in my game with no normal map because "the engine can handle it, so shrug" I still don't want all of that wasted data in my project or my builds.

Another question I have to ask myself is: does that system need the power of a not-yet-released console sitting behind it just to function? Other questions are "so how much memory and storage are being chewed up to store all that data" and "does this mean gamers will expect even more?" People have pointed out that you can just scan stuff now, and that's true, but how many games are made wholly or even primarily out of stuff that really exists to be scanned? How much of that stuff do you have physical access to? And then how much of it would require no modification?

Still, exciting stuff, and do keep in mind that it's a demo. Some of my first impressions were things like "why are they showing gameplay the same as we saw on the PS3?" and "this interactivity isn't well designed!" But when I pushed that stuff aside, the tools on display here have a lot of promise.

 

Technically probably no (you do need compute) ... at that quality level, probably yes

For visual assets weight, it's probably texture all the way down (even the mesh), compression and lod should be the same somewhat thing, which mean it would scale very well by doing texture sampling trick. Virtual texture all the way down.

 

Yeah, it's not like indies would have the resources to generate millions and millions of polygons, but the workflow benefits of not having to generate LODs and normal maps? Amazing.

I believe the answer is no. This is from the top of my head, so grains of salt and all, but I think they said that this demo is designed so that it could scale down to say, a PS4, although obviously much worse looking.

 

Unity has said they're working on their own in-house realtime GI solution. And UE's GI isn't coming till 2021-2022. I'm sure Unity won't be too far behind with their own.

 

I think the problem is unity does not properly advertise their tools and don't demonstrate their power. the implications of dots and ui are much greater than some graphics change that makes rocks look better when close up.

being able to create high performance ui based apps for every device is pretty big. so is producing multi threaded vectorised code with ease. much more than ue 5 would supposedly have

 

god it won't let me fix typos

 

Those are not ready though and actually I would argue, they have already advertised them a whole bunch, considering how not ready they are.

 

Those things, and the stuff UE is now showing off, are all highly dependent on what you're doing. Plenty of games I see made with Unity wouldn't benefit from that stuff. Plenty of those games also wouldn't benefit from DOTS.

It's definitely a "best tool for the job" thing. Both engines are pretty great tools.

 

I would disagree
DOTS would benefit every game. there are no dependencies so you don't have to drag gameobects to fill public fields. there are no null reference exceptions. GetSingleton<> is just so convenient, you don't need to make singletons anymore with

Code (CSharp):

1. public class MyClass : MonoBehaviour {
2.     public static MyClass singleton;
3.    
4.     private void Awake() {
5.         singleton = this;
6.     }
7. }

not to mention that unity is all in on dots, so either way if you want the latest animation, netcode, physics, audio, editor workflow, etc you need to use dots

 

Visual scripting is made for people who can't program. It is largely a crutch that is only really useful in a very few areas.

 

Them being all in on it doesn't mean it's done though. The pre-DOTS systems are stable-ish and most of DOTS is still in preview.

 

I think it's pretty awesome for mapping out anything that can be represented by a graph. I don't necessarily want to do the whole "do a for loop in Visual Scripting", but an FSM or behavior tree? Sign me up.

 

no, i was just saying that dots is not a specialised tool for a specialised task, but a replacement for the whole api

anyway it'll probably be production ready by 2021 when ue5 comes out, so it would be nice to see how the two engines would fare in the next generation. now would be a good time to learn dots if you plan to release a game by that time.

 

It is though. It makes simple cases much more annoying to code without much benefit.

 

To go back to Unreal asking for nothing up to 1million $ in revenue per title, if Unity could F***ing sort out their licensing kerfuffle, so that me, a Pro subscriber, can have a freelancer Audio guy work on my game with the Unity editor, without me having to shell out 1800$ for a license that will only be used for the couple of months the Audio guy will use it, THAT WOULD BE KINDA NICE.

I've been saying it a lot, but that's kinda the opposite of democratizing development.

 

well.. that is if you're not familiar with the api. to be fair, there are literally 0 learning resources.
the two are not that dissimilar

Code (CSharp):

1. public class DOTSClass : SystemBase {
2.    
3.     // Start is called before the first frame update
4.     protected override void OnCreate() {
5.        
6.     }
7.  
8.     // Update is called once per frame
9.     protected override void OnUpdate() {
10.        
11.     }
12. }

OnUpdate = Update()
OnCreate = Start()
GetComponentDataFromEntity<>()[entity] = gameobject.GetComponent<>();

 

We generally have one manager script that handles everything per scene and 3-4 helper scripts, what would I gain from this?

 

well, what's your game? would you mind sharing a source snippet so i can prototype an ecs equivalent?
you get better animation, better audio, better netcode, better physics, better editor workflow.

you could get your manager script with GetSingleton<ManagerScript>();
there's also RequireSingletonForUpdate<ManagerScript>(); which would make it so update only ever runs when the singleton is present
you can change the order in which the scripts are ran

Code (CSharp):

1. [UpdateBefore(typeof(HelperScript))]
2. public class DOTSClass : SystemBase {
3.    
4.     // Start is called before the first frame update
5.     protected override void OnCreate() {
6.         RequireSingletonForUpdate<Manager>();
7.     }
8.  
9.     // Update is called once per frame
10.     protected override void OnUpdate() {
11.        
12.     }
13. }

 

They are releasing a Unity SDK for this so it can still be used in unity.

 

It's a point and click adventure game. There's a game logic script, and a bunch of scripts that handle movement and pathfinding for NPCs and the main character. This is not the time or place for me to share source code, sorry.

Netcode and physics do not apply. Not sure what you mean by better audio. Does DOTS allow me to bypass fmod? AFAIK the audio mixing stuff were in preview last time I checked them out.

 

Huh? This is going to save people what... 15 seconds at most?

Ok then. What benefit would I get for switching my game to use DOTS right now? If it "would benefit every game" then you ought to have some concrete examples off the top of your head as to why things would suddenly be noticably better for my players if I were to put in the effort this second.

And that's not considering that I'd have to learn a bunch of new stuff, and possibly change how we're doing things moving forward.

 

For those arguing that your game would require large amount of storage space because as we know hi-poly meshes. However, as said by folks at Epic games in this video, UE5 will scale down from mobile all way up to high end PC. The same way you can tell your virtual texture program to use X amount of memory budget.
Now imagine you say "UE5, please build my game at 25% of original geometry/texture resolution for this platform", not just it would take less storage space but also run a lot faster cause there's less data to be streamed from disk! And on top of that the premise of developing the content once and the engine scale it for you everywhere, saving production costs/time is just enough for anyone to make the switch.

 

This doesn't address filesize concerns at all. Even if you reduce the statue model, the one they use as a reference, down to 25% of uncompressed, that's still five hundred and twelve megabytes for that one asset. There are a lot of storage considerations to make here. We've seen literally nothing of the scaling they're talking about.

 

yes, dot's audio is extremely ow level. you make your own channels, sample rate, attenuation, stereo. would wait until 2021 if this is a real game, its worthwhile learning though. there's no audioplayer.play("sound"), that's the caveat. i have 7 200+ line scripts just for playing a single sound that's at a certain location. the general issue i've encountered with dots right now is not that its in preview and thus unstable or feature poor, but that there are no standard assets and so you just have to work on the absolute lowest level. its just soo mind bogglingly intricate.. unless you use Project Tiny. they have their own dots audio and animation systems that's allot more on the level of AudioPlayer.play();

ok i retract my position, but when unity releases a higher level api scripts in 2021. then it would benefit everyone. if your game is mobile or something, you stand to benefit allot from battery consumption / heat. also you could achieve megabyte size builds if that means anything to your players. really depends on your game.

mecanim is pretty expensive and you could gain allot if animations are the focus of your game, there are procedurally animated animations that blend better with their environment

 

"UpdateBefore" is pretty much a heresy as far as code organization is considered, as it makes execution order non-obvious. It is not as bad as specifying classes, function call and variables using a string (Python/SqlAlchemy does that), but approaching that. Existence of such directive is a very good reason to avoid Dots.

Not "when", "if".

 

I think DOTS is cool and it has a host of benefits, but for beginners at least, figuring out how to write code will be more difficult than ever. I am skeptical of the decision to migrate to a less approachable programming architecture largely on the basis of performance, and as an asset store developer it makes me uneasy. I think Unity are banking on the success of their visual scripting tools and if that's the case I definitely don't think it's a good strategy.

Not to mention all the learning resources out there where you can google for the solution to just about anything will become out of date.

Unity should instead focus on the nuts and bolts of game development in general, doing away with unnecessary and unwieldy phenomenon such as light probes and LOD baking, polishing and stabilizing core features of the engine, and generally taking a breath or two before introducing some other new or overhauled feature that will likely be gone in a few years while still sporting significant issues.

 

I have seen presentations. I wasn't impressed actually, and for the record, I used to program my own VST synth instruments, but whatever it is better than what we have now.

See that's a big no-no for me though. 200 lines to play a single sound?

That's also a big no-no for me.

See the answer here though is not to go DOTS, it's to use the Legacy Animation system, which for simple animations is so much faster than Mecanim, it's kinda insane. If your bar for why we should use DOTS is : it's faster than Mecanim, then that's a so low bar, I'd argue it's not worth it.

Also, with all that said: I'm sort of playing devil's advocate here. I bet if I invested heavily in DOTS there would be wins there (major or minor, I don't know). As it is right now though, I will not bother with DOTS, until it's ready. I have enough headaches with "production ready" parts of Unity, I will not add "not ready" parts to my workflow, especially not for not really tangible benefits and a ton of work on my side to transition.

 

Honestly, there's few things I'd like more than a "spring cleaning" major release that does this sort of thing. I want an entire 202X cycle dedicated just to workflow improvements and stabilizing features. I know that's what they're talking about doing for the next few updates, but I dunno how well it'll stick.

 

One important thing to keep in mind with the DOTS visual scripting system is that it's able to generate code. In my opinion this has the potential to make it one of the best learning resources for DOTS. Whether it has any use beyond that we'll have to wait and see.

 

don't think there would ever be a case where you would need to access a singleton in the awake.. you mostly need singletons for update. OnCreate is used allot less often than Start(), GetComponentFromEntity() in update is not bad for performance compared to GetComponent<>().

from what i understand, the general consensus was singletons = bad design. but for ecs, it was quickly agreed that singletons = in actually very useful. overwatch uses a crap ton of singletons from what i understand.

don't see whats wrong with update before / after, but there is an alternative which is updating in different groups

Code (CSharp):

1. [UpdateInGroup(typeof(InitializationSystemGroup))]
2. public class abc : SystemBase {
3.     protected override void OnUpdate() {
4.        
5.     }
6. }

simulation is the default group. initialisation happens before simulation, and presentation happens after simulation

I am sure that it is completely certain that unity will implement some higher level api on the level of Animator.play("walking"); and they've talked about it being in the works on multiple occasions. then it would be epic. anyway since we're talking about ue5, i think its fair to bring up an idealised future state of dots. dots audio and dots animation is not out in the open in the package manager anyway, you have to enter a special cheatcode for it to be revealed to you. but anyway i'd still recommend you learn it now, just as i would recommend you learn ue5 if it were for preview now

 

That's a good thing. But what I expect to see is a significant part of the user base (maybe even a large part) who find the visual scripting too cumbersome to use, and the DOTS approach far less comprehensible than the component-based stuff. This has the potential to send a lot of people off to greener pastures.

 

Which is problematic when you consider that the roadmap calls for it to leave preview with Unity 2020.1.

http://www.cgchannel.com/2019/09/unity-technologies-unveils-unity-2020-product-roadmap/

 

Sure, sure, like how I've spent all this time learning how to get good results out of Enlighten, which was obviously such a good way to spend my time.

Unity is fickle.

Don't spend time on anything that:

1. Isn't ready.
2. You don't need right now.

Or you may be wasting your time.

 

Couldn't agree more.